Implicit conversion from float -46.254980079681275 to int loses precision